cancel
Showing results for 
Search instead for 
Did you mean: 

Issues with AP250 and HiveOS 8.1r2?

Issues with AP250 and HiveOS 8.1r2?

Matt4
New Contributor

Just after the first of the year I updated all (59) of my AP250’s from 7.0r2 to 8.1r2. It was a major struggle getting all the AP’s updated, which should have tipped me off to stop but I powered through. In the days following we had numerous reports of tenants having wifi issues. The main issue we had was a user would be connected working perfectly then suddenly, with device still connected, they could not access anything on the internet, then a suddenly as it quit it began working again.

There were also issues with the Hive being unresponsive, reporting incorrect data from AP’s, turning on bonjour gateway in the delta when it was turned off (this was only in the delta, a complete config would have it off). I worked extensively with tier 3 support, we downgraded firmware, upgraded graded firmware, meticulously combed through the network profile making some tweaks to the radio profiles, data rates, QoS, and firewall settings. We rebalances 2.4GHz radios to fix some interference issues. Nothing seemed to help the issues. After almost three months of working with support and getting nowhere I decided to roll back to 7.0r2 and, knock on wood, things have been running smoothly again.

I should mention that our environment is a little unique. We are an off-campus student housing building, so basically apartments. We offer our tenants “free” wifi, so all of the devices on the network are BYOD, which we have no control over. I should also mention that the main issues was sporadic, one tenant would have an issue while their friend sitting right next to them would be working perfectly fine. We had reports of issues from all device types, iPhone’s, android phones, macs, windows pc’s, rokus, apple tvs, fire sticks.

All of that to ask. Has anyone else had issues with HiveOS 8 versions?

1 ACCEPTED SOLUTION

nlowe
New Contributor III

Hi Matthew,

 

I work in the UK Tier 3 team at Aerohive, one of a couple of people who handle support escalations in the EMEA/APAC regions with Aerohive's engineering.

 

We are aware of one behavioural difference present in HiveOS 8.2r1a and HiveOS 8.3r2 and not HiveOS 7.x whereby a dynamic OFDM sensitivity threshold now present, used to decide whether to demodulate or not for RX, can cause issues with some stations where their transmission is received at low power.

 

The impact is that you may observe things like RTS/CTS not taking place where this is received below the dynamic threshold, with sticky clients at the edge of a cell this can be challenging.

 

This threshold is inevitably a tradeoff of various factors to ensure the best airtime efficiency but we are looking to see if this logic can be enhanced to better mitigate this side effect.

 

Symptomatically, this can match what you have described.

 

In HiveOS 8.2r1a and 8.3r2 as far as CFDs go:

 

CFD-2973 is unlikely to be related as it pertains to an explicit rejection of auth/assoc with some clients. This was not an intermittent issue and was an interoperability concern.

 

CFD-2973 pertains to an interoperability issue in high density setups where there are a mix of legacy 802.11b CCK and 802.11n OFDM rates for the basic rates. The fix constrains the management frame behaviour so that only the CCK rates are used.

 

CFD-3076 made changes to the CCA threshold, setting it more appropriately according to the region configured as opposed to the same value across all regions. This can improve connectivity in noisy, interference prone environments.

 

CFD-3039 resolved a TX queuing issue where the radio was switching between servicing clients and performing background scanning. This could cause packet loss with some clients.

 

Please do test with 8.3r2 where possible or 8.2r1a otherwise.

 

Thanks,

 

Nick

View solution in original post

15 REPLIES 15

AnonymousM
Valued Contributor II

Matthew,

 

Similar issues here, but with basically any AP (majority were AP230s). We are a medium sized MSP and have customers all across the country and when we scheduled maintenance to move firmware to 8.x (moved to the KRACK patch HiveOS specifically - 8.2r1a I believe), we heard almost identical reported issues from NYC to California and everywhere in between.

 

It was an absolute train wreck as is that HiveOS version (in my eyes). We've been stuck using 6.5r4 since it's release as it's been the code we've had the most luck with stability wise and we've encountered issues on the newer 6.5r5-9 HiveOSs as well.

 

 

GTM-P2G8KFN